<p>A method of obtaining a motion vector in moving images is based on evaluation of computational results of searches made by matching a target template and a group of pixel blocks each having a same size as the target template. The steps include: searching in a wide area and obtaining a reference vector to indicate an overall image movement (S1); searching in narrow areas centered about the reference vector for all templates in the target image, and obtaining a displacement vector for each template in relation to the reference vector (S2); and summing the reference vector and the displacement vector for each template, and assigning a computed result as a motion vector in each template (S3). <IMAGE></p>